Deep Learning in Production

Explore deep learning deployment in production environments through this 44-minute conference talk by Adam Gibson, founder of Skymind.io and creator of Deeplearning4j. Gain insights into the Deeplearning4j framework, the most popular open-source deep learning solution for the JVM. Discover effective strategies for implementing deep learning algorithms in various applications, from computer vision and fraud detection to network intrusion and self-driving cars. Learn about research clusters, HPC video transcoding, big data hybrid clusters, Mesos, Spark, and inference techniques. Delve into topics such as TensorRT, latency hiding, prefetch buffers, Flux, and the deep learning ecosystem. Understand the importance of tensors, GPU cluster memory management, and pretrained models. Acquire knowledge on low latency solutions and system information crucial for successful deep learning deployment in production environments.

Intro
Defining Production
Research Clusters
HPC Video Transcoding
Big Data Hybrid Cluster
Mesos
Spark
Inference
TensorRT
Hiding
Latency Hiding
Prefetch Buffer
Flux
Latency
Deep Learning Ecosystem
Tensors
James Weaver
GPU Cluster Memory Management
GPU Memory Management
Kaggle
Data Developers
NASA
Low Latency
Pretrained Models
System Information